AI Agents

An org of agents that runs the work

Muntri's agents are organized like a team: supervisors that direct the work, repeatable processes that execute it, and per-record assistants that know your data. You decide how far each one acts.

app.muntri.com/agents
Muntri agent org chart: master, process and context agents
The agent org

Three tiers, one hierarchy

Agents are arranged in an org chart you can drill into, from the supervisor down to the process it runs and the record it acts on.

Supervise

Master agents

Master agents oversee the work for an area and hand off to the right process, so the whole motion stays coordinated.

Execute

Process agents

Versioned, visual workflows, drag-and-drop, that execute a repeatable motion: qualify a lead, work a deal, run a renewal.

Ground

Context agents

The per-user and per-account "Ask {name}" assistants, grounded in that record's real data. This is Ask Muntri.

Supervise

Master agents run the motion

A master agent supervises the work for an area of the business and routes each case to the process that should handle it, with bounded per-account overrides on the steps you flag.

  • Supervise a whole area, not a single task
  • Hand off to the right process agent
  • Per-account overrides on owner-flagged steps
  • Drill into the org chart from supervisor to step
app.muntri.com/agents
Muntri agent org chart with master and process agents
Execute

Process agents are visual workflows

Each process agent is a versioned, drag-and-drop workflow over one execution engine: triggers, conditions, actions, AI and branches, with parallel and fan-in. Change it on a canvas, and every version is tracked.

  • Drag-and-drop canvas, one execution engine
  • Versioned, so you can see and roll back changes
  • Triggers, conditions, actions, AI and branches
  • A repeatable motion per process
app.muntri.com/automate/workflows
Muntri visual workflow canvas for a process agent
Ground

Context agents know each record

Every user and every account gets its own "Ask {name}" assistant, grounded in that record's real data. Ask your data in plain language and get cited answers, from the agent that actually knows the account.

  • A per-user and per-account assistant
  • Grounded in that record's real data
  • Plain-language questions, cited answers
  • This is Ask Muntri
app.muntri.com/accounts
Muntri context agent grounded in an account record

Per-action autonomy

You decide how far agents act

Every action carries a governance tier, so agents run on your terms, action by action.

Auto

Act on its own

The agent runs the action without a hand-off, for the work you're happy to fully automate.

Approve

Human in the loop

The agent pauses and asks for approval first, then runs the same step once you sign off.

Advisory

Recommend only

The agent suggests the action and logs it, but never executes, so you stay fully in control.

Agent Hub

Build your own, or start from a template

Build custom agents on your own data, or start from a marketplace of templates and adapt them. Every agent gets its own avatar, so the org feels like a team.

  • Custom agents built on your data
  • A marketplace of ready-made templates
  • An avatar for every named agent
  • Master, process and context, in one hub
app.muntri.com/agents
Muntri Agent Hub with custom and marketplace agents
Grounded and governed

Real data, real guardrails

Agents act on your real data, within the guardrails and budgets you control, and route to the AI model you choose, at vendor token cost. Grounded in your record, governed by your rules.

Bring your own AI

Put an org of agents to work

We'll stand up a workspace on your own data and show the agent org running your motion, on your terms.

Explore the platform